1Ghz Wilkinson Power Divider Implementation

[sapphire]

I wonder whether the 1Ghz Wilkinson Power Divider could be implemented inside the chip?

Because the quarter-wave line of 1Ghz could be as long as 23 mm, the Wilkinson Power divider would occupy a lot of chip area.

Thanks!

 

[sinatra]

AT 1GHz you will have the following parameters:

1-wavelenght in air=30cm
2-Quarter wavelenght=7.5cm
3-Quarter wavelenght of a microstrip in a silicon or indiun phosphide substrate=2.1cm
If you make a zig-zag you can probably reduce the lenght by a factor of maybe 5.
Then the quarter wavelenght would be=4mm.

I think it is still too long, you are probably not yet in business...
The frequency seems too low for an implementation of a Wilkinson splitter in a 2 by 2 mm^2 area for example.
S:

 

[sapphire]

I love your reply. Very helpful.

I wonder what kind of topology should I use for a 1Ghz power divider in GaAs. BTW, I am still a student.

Thanks a lot!

 

[sinatra]

You can build the same type of combiner, i.e., Wilkinson combiner/splitter, but using lumped elements.
Every quarterwavelenght transformer is replaced by a low pass lc circuit (Cp1 Ls Cp2).
The 100 ohms resistor will still be there.

Using this idea you can go further and build any hybrid coupler, etc.. and in this way you would need no resistor.

I would personally advise you to build a branchline hybrid using lumped elements . You can probably build very good capacitors in your process. You will need to build spiral inductors.

You should be able to simulate the structure so that you can optimize and tune it in such a way that the parasitic capacitances of the inductors are absorved into the circuit, you will probably need to reduce the lumped capacitors during this optimization.

S.

 

[sapphire]

Thank you very much!